Download
1 / 35
350 likes | 476 Views
ITT - Environment Agency Challenges. Dr Sean Longfield Lead Scientist, Flood & Coastal Risk Management Research 9 November 2017. The Environment Agency. we work as part of the Defra group (Department for Environment , Food & Rural Affairs )
E N D
ITT - Environment Agency Challenges Dr Sean Longfield Lead Scientist, Flood & Coastal Risk Management Research 9 November 2017
The Environment Agency • we work as part of the Defra group (Department for Environment, Food & Rural Affairs) • a cleaner, healthier environment which benefits people and the economy • a nation better protected against natural threats and hazards, with strong response and recovery capabilities
Research at the Environment Agency • Flood and coastal risk management • Small catchments • River flow trends • Air, land and water • Nitrate vulnerable zones (NVZs) • Climate change and resource efficiency
Flood and Coastal Risk Management Research Joint Defra/EA Flood and Coastal Risk Management Research and Development Programme….The Joint Programme Policy, Strategy and Investment Asset Management Incident Management and Modelling To find out more, visit:http://evidence.environment-agency.gov.uk/FCERM/
1. Nitrate Vulnerable Zones (NVZs) Nitrate Vulnerable Zones (NVZs) are areas designated as being at risk from agricultural nitrate pollution they include about 58% of land in England borehole network across England (points) what might the nitrate concentration be in groundwater between these points? currently use Kriging to interpolate between points this doesn’t take account of underlying geology, which can have a large impact need a method of interpolation that can account for geology
1. Nitrate Vulnerable Zones (NVZs) disjunctive Kriging? not a uniform geographical distribution of boreholes boreholes at different depths lots of data from the 1980s onwards approx. 3,500 borehole sites Challenge:develop a new method for interpolating nitrate concentrations in groundwater from borehole data in England that takes account of underlying geology
2. Statistical Flood Frequency Estimation:Fundamentals estimate a flood peak (m3s-1) of a given rarity, the T-year flood where T expresses event rarity as a return period in years the return period of a given flood is the average interval between floods of this magnitude or greater gauged records are rarely long enough to allow direct estimation we often need to estimate at ungauged sites
2. Statistical Flood Frequency Estimation:Data Catchment Descriptors POT AMAX
2. Statistical Flood Frequency EstimationMedian annual flood (QMED) • QMED, 2-year return period flood • exceeded on average “every other year” – the index flood • With gauged data • median of AMAX data • from POT data • Without gauged data • from catchment descriptors
2. Statistical Flood Frequency EstimationSingle site fit a distribution to AMAX data (GLO, GEV, log-normal, kappa…) estimate parameters using L-moments derive a growth curve (scaled to 1 at QMED) derive a flood frequency curve (product of QMED and growth curve)
2. Statistical Flood Frequency EstimationEnhanced single site • pooling group created based on hydrological similarity • derive pooled growth curve • more weight at the site of interest • select suitable distribution (GLO, GEV, log-normal, kappa…) • estimate parameters using pooled L-moments • derive a flood frequency curve
2. Statistical Flood Frequency EstimationUngauged site • create pooling group based on catchment descriptors • derive pooled growth curve • Select suitable distribution (GLO, GEV, log-normal, kappa…) • estimate parameters using pooled L-moments • derive a flood frequency curve
2. Changing flood return periods of over time • Flood frequency estimation • estimation of peak river flow of a given frequency (rarity) • we know that design flood estimates for a site can vary dependant on the time period chosen for analysis • There are well known, ‘flood-rich’ and ‘flood-poor’ periods over the past 100-years or so.
2. Changing flood return periods of over time Challenges: Can we develop an indicator, or measure for river gauging stations that tell us how sensitive a flood record is to the time period chosen? How does that vary across the country? How sensitive are design flow estimates at individual sites to ‘extreme’ events – can one big event change design flows significantly, or is the site relatively insensitive to high magnitude events?
2. Clustering of floods events • There is a perception that flooding is happening more often (rightly or wrongly) • This raises some interesting questions/challenges: • how can we detect significant clusters of flood events? • are there any statistical methods that can be used to describe ‘clustering’ of flood events (say from POT data), and how that might have changed over time? • does flooding happen more now that in the past? • if a 10-year event has occurred one year, what is the probability that a 100-year event might happen the next year, or in the next 5-years…
2. River flow trends we have many decades of river flow (and water level) data at point locations in the UK we often try to look for evidence of river response to climate change (and/or land use change) in these records Challenges: what is the best data to use to look for temporal trends – AMAX, POT, monthly, daily, 15-minute, and why? how many years will it be before we can say with confidence that there is or is not a trend in river flows (daily, AMAX, POT). how much river flow data do we need before we can say if climate change is having an impact or not?
3. Small catchment hydrology catchments less than 25km2 usually no gauged data high uncertainty with these methods Challenge:Could you take the data we have available and devise a way of estimating low probability flood flows (0.5% - 0.01%+ chance of occurring in a given year)? Alternative Challenge:If we were to look at the available data and methods today, would we still apply regression analysis? Have new techniques emerged that could reduce uncertainty? Would we do the same – but differently? Variation on the above – What if we limit ourselves to “open data”
3. Small catchment hydrology Can methods be made widely available? i.e. simple spreadsheet or web-based execution as opposed to requiring complex statistical software? Can methods rely solely on open-data? Methods would probably need to describe the potential flows against their probability of occurring – and uncertainty As a bonus – can hydrograph shapes be extrapolated from catchment descriptors (geography) and existing flow records?
3. Small catchment hydrology • Potentially useful data: • Flow and level records (15 minute) (some open) • Conversions between level and flows (some open) • FEH Catchment descriptors (not open) • Rainfall radar (spatially continuous) archives (not open) • Rainfall gauges (point sources) (part open) • Potential alternative catchment descriptors: • BGS Geological data (Partly open) • Various continuous DTMs (OS/USGS/Global DTM) (various – open to not open) • EA LiDAR (has gaps) (open) • Open environmental mapping (OS, OSM, Natural England, Environment Agency)
